Fan PID about t-962-improvements HOT 13 CLOSED

The PID most definitely needs to be tuned to avoid the annoying oscillation. It completely freaks out for a short while when at the top of the reflow. Other implementations (ESTechnical comes to mind) uses two separate PID-tunings for the (mostly) heating part and (mostly) cooling part. Not sure if it makes sense if the loop is properly tuned when I arranged cooling and heating as one continuous output. This needs more thought and experimenting, but when I graphed the temperature around reflow I did not really notice it staying up there for too long (as the oven has some issues reaching the temperature in time, so my total reflow time looked reasonable). Maybe you actually reached the setpoint earlier as you have more power available in the A-version? In that case we might need different tuning for different oven sizes, something I seriously doubt is in the original firmware.

from t-962-improvements.

That makes a lot of sense. I tested it using the "low temperature" profile so it probably did not need to work so hard to get up to temperature and would not drop as fast once at the peak. I'll test it again with a higher temp profile. I only used the lower temp for testing to avoid the hot exhaust on the bottom burning the top of my desk next to my PC!

from t-962-improvements.

I've done exactly the same when testing most cycles! :) That's also where you can notice the oscillation during heating a bit more. I have successfully reflowed a few boards with the low-temp paste with this profile without noticing any de-wetting so it seems to be within the margin of error. You can see that I did not pick the absolute highest temperature as max reflow temp in the profile either. For higher-temp lead-free the timing is much more critical.

from t-962-improvements.

I have changed the way the PID regulator gets its setpoint now which should improve this scenario (the PID now knows just when the reflow peak is over and will adjust much faster). By giving it a step-wise input it actually follows the curve much better (as it now gets to know where we want to be in 10 seconds).

from t-962-improvements.

The new PID seems better with regard to quickly dropping after the peak.

I noticed that during the warming part of the profile, the actual temp is consistently a few degrees cooler than the profile. Probably not a big deal if it is just a time lag if the entire profile is delayed the same amount.

from t-962-improvements.

Great! Well the lag at the top end is a bit difficult. It seems to be thermal lag in the thermocouples (they take a while to reach the correct output). If you measure the temperature of the air with a DMM thermocouple, is it still low?

Regarding the entire profile lagging, yes the PID definitely needs more tuning to be a bit more aggressive (but not too aggressive). I haven't had time to run through enough cycles yet, and it seems that auto-tune is not very useful. The resulting reflowed boards looks really nice though with the lead-free profile, but the regulation should be improved anyway.

from t-962-improvements.

I measure the oven temperature with a thermocouple and a DMM where the thermocouple wire is held down to the bottom of the sliding drawer and the probe tip is centered between the two oven probes.

The DMM shows readings that are fairly accurate at 140 C. At 240 C, the DMM shows a temperature that is initially about 10 C too high, but after a few minutes, the DMM reading slowly drops until it is only about 3 C too high. I'm not sure what would cause the DMM readings to change while the oven's 2 thermocouples do not change. The cold junction readings do not change during this time.

On one hand, it makes me wonder how useful calibration would be if the oven's hardware design does not permit repeatable temperatures any closer than that. On the other hand, calibration may be useful to easily shift profiles up and down in temperature without having to re-enter a custom profile.

from t-962-improvements.

I have seen similar readouts and the only explanation I can think of is
that the thermocouples have some of the heat shunted away into the
mounting which makes them a bit slow when the temperature is hot. I have
similar issues with some of my DMM thermocouples and I've been playing
around with a few different brands trying to find the fastest ones
(smallest diameter wire and spot weld makes for a fast thermocouple). So
far the best ones have been these ones:
http://www.amazon.com/gp/product/B0087ZR81O/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1
. Unfortunately one of them developed a nasty non-linearity after I
accidentally kinked the wire, so it would OK up to roughly 180C and then
read 10-20C too low higher than that, consistently.. Also they claim
400C max temp but you really need to remove the heatshrink tubing for
oven use. The connectors I got were the removable ones with screw
terminals so they were super simple to strip to bare wires and install.

from t-962-improvements.

Was there any progress on improving the PID tuning.

I've just flashed got my T-962 and flashed the latest firmware, and have run the oven a few times to test how well the Actual values track the Set values, and I'm generally seeing that the Actual temperature is about 5 degrees below the Set value, when the temperature ramps up.

The only time I see a slight overshoot is at the very peak of the reflow.

It looks like the PID needs to have a bit more gain (but I'm not an expert in PID).

I wonder if it would be worth adding the PID tuning values to one of the menus.
I'll probably for the code and have a go at doing this myself.
I've compiled and reflashed the existing code, and I'm familiar with embedded programming - however I'm not sure how easy its going to be to make the setting menu scroll up and down to accomodate new settings ;-)

from t-962-improvements.

+1 to able to tune FAN PID too.

from t-962-improvements.

@waldo3000

I've been doing a load of test with 5 different thermocouples, either connected to the oven, or to a DMM with thermocouple input.

But consistency between the thermocouples is very poor.

I have given up using the opamp sensing electronics on the control board and I'm using Max31855 interfaces, but this doesn't seem to help a lot.

Ive now ordered another 5 themocouples from eBay, and I have 4 Max31855 interfaces, which I'm going to connect to a separate microcontroller board (STM32) and I'm going to see what results I get using a separate system to read the values.

I'v also modified by fork of the firmware so I can use offset and gain on the Max31855 inputs, as that feature was not available in the main version (here)

BTW. It should not be too difficult to put PID parameters into the Settings, as I already had to do this so that all 4 of the possible Max31855 inputs can have these corrects applied to them, and it was not too hard to add the the codebase

from t-962-improvements.

Issue cleanup

from t-962-improvements.

In the end, going to all this trouble didnt really help, as I still got loads of hotspots in the oven, which made it unusable for me.

1 year on, I'm now starting to think about fitting a proper convection fan, but that's an engineering job not a electronics or programming problem

from t-962-improvements.